On Tue, Apr 17, 2018 at 07:11:34PM +0200, Daniel Kiper wrote: > CC-ing Vladimir. > > On Mon, Apr 16, 2018 at 06:05:04PM +0800, Michael Chang wrote: > > The next_buf is the offset to the next cached block rounded to the size of > > bufio->block_size. However the calculation needs the block_size to be in > > power > > of 2 is not always valid. As an example, files with smaller size than > > block_size will have the block_size leveled to the size of file which can be > > set arbitrary value. > > > > This patch fixes the next_buf calculation to accept any integers. > > > > Signed-off-by: Michael Chang <mch...@suse.com> > > --- > > grub-core/io/bufio.c | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/grub-core/io/bufio.c b/grub-core/io/bufio.c > > index 22438277d..d0b0f71b6 100644 > > --- a/grub-core/io/bufio.c > > +++ b/grub-core/io/bufio.c > > @@ -132,7 +132,7 @@ grub_bufio_read (grub_file_t file, char *buf, > > grub_size_t len) > > return res; > > > > /* Need to read some more. */ > > - next_buf = (file->offset + res + len - 1) & ~((grub_off_t) > > bufio->block_size - 1); > > + next_buf = (grub_divmod64 (file->offset + res + len - 1, > > bufio->block_size, NULL)) * bufio->block_size; > > Should not you fix this in grub_bufio_open()? I think that > you should look for closest power of 2 in it.
Of course here we can round up or down the bufio->block_size to meet power of 2. The down side of round-down is inefficient for small files as it can't cache entire file in one go. The round-up will have to allocate (slightly) larger than needed buffer to hold small files which is not a big deal ... If you insist, I can work a new patch to round up the block_size in grub_bufio_open, just let me know. Thanks, Michael > > Daniel _______________________________________________ Grub-devel mailing list Grub-devel@gnu.org https://lists.gnu.org/mailman/listinfo/grub-devel